Presentation on screen can be merely decorative, or it can reveal and/or manipulate the nature of the ideas presented. Thinking along these lines, how could 3D presentation help to communicate meaning ?

The normal tag cloud looks like this (as to the left) :

A 3D version would look something like this (font size would indicate number of tags) :

When you click on, say 'politics' everything would swirl around, 'politics' would adopt the largest font size, and words to which there were links would adopt relative font sizes - with connections shown. All the rest would be shaded into 'background' ... looking like this :

This is just a very simple sketch. A fully developed cloud, with a lot of words and a lot of links, would have much more subtle gradations, and probably many more links (the 'background' might be arranged by numbers of links to the words in the central, principal ring of links).

I would argue that such a presentation technique would illustrate something about the conceptual structure of the cloud - and would not be just a decorative feature.

As it happens, software along these lines might have great potential for developing students' vocabulary - networks of related words (perhaps with definitions produced when you double click, or something ?)